The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest data from multiple external sources, which constitutes a potential attack surface for indirect prompt injection if those sources are compromised.
Ingestion points: The skill reads input from short-form-video-script, brand-profile, design-and-templates, and voice-builder to inform the editing plan.
Boundary markers: The workflow utilizes a clear human-in-the-loop boundary where the agent provides plans and specifications, but the human user is responsible for final execution and approval in the CapCut application.
Capability inventory: The agent's capabilities are limited to generating instructional plans and specifications (e.g., cut plans, caption styles, export specs) and it does not have autonomous tools to modify the file system or execute shell commands.
Sanitization: Safety is maintained through manual human review of the agent's output before it is applied to the video project.
[SAFE]: The skill includes explicit security and privacy guidance, such as advising users to edit confidential or NDA-restricted material locally without cloud synchronization to maintain data ownership and comply with client agreements.

capcut